1

Conducting randomised controlled trials with older people with dementia

in long-term care: challenges and lessons learnt

ABSTRACT

The characteristics of older people with dementia and the long-term care environment can

make conducting research a challenge and, as such, this population and setting are often

understudied, particularly in terms of clinical or randomised controlled trials. This paper

provides a critical discussion of some of the difficulties faced whilst implementing a

randomised controlled trial exploring the effect of a live music program on the behaviour of

older people with dementia in long-term care. A discussion of how these challenges were

addressed is presented to aid investigators planning the design of similar research and help

encourage a proactive approach in dealing with research-related challenges right from project

conception. The article is structured according to the three principles of a randomised

controlled trial in order to keep experimental rigour at the forefront of this research area.

Key words: cross-over design; dementia; methodology; randomised controlled trial; research

design.

2

INTRODUCTION

Dementia is a clinical syndrome with a number of different causes. It is characterised by the

decline of a person’s cognitive functioning, affecting skills in language, memory and

perception¹ and can lead to an increase in challenging behaviours such as agitation, aggression

and wandering.2 Such cognitive and behavioural changes can make it more time consuming

for long-term care (LTC) facility staff to provide care for people with dementia3 and can lead

to greater levels of stress.4 This can then have implications for the retention and recruitment

of staff but also, more importantly, may negatively affect the quality of care provided to the

person with dementia.5 Given these concerns, coupled with statistics that show dementia to be

an increasingly prevelant condition,6 there is a need to understand how these challenging

behaviours may be ameliorated. Such research may be particularly necessary in light of the

increasing number of older people living in LTC in many countries.7,8 However, undertaking

research with people with dementia and then more generally in LTC is often fraught with

challenges.8,9 For instance, the level of cognitive decline and the presence of agitated

behaviours may affect or compound consistent intervention implementation10 and make it

difficult for the person with dementia to complete outcome measures.11 In addition, acute

illness or death also makes maintaining a large enough sample size an ongoing struggle.12

When examining the challenges encountered in the LTC environment, inflexible facility

routines, policies and practices, 7,9 staff non-compliance with research protocols, 7 high levels

of staff turnover, and reduced staff-to-resident ratios9 have all been found to make research

cumbersome. As a result, research conducted within LTC with both frail older people and,

specifically, those with dementia has generally been understudied.8,13 This has particularly

been the case for clinical or RCTs13,14 owing to the required rigour and extent of control over

extraneous variables needed.

3

The adoption of psychosocial treatments for dementia has received increasing research

interest and the therapeutic use of music has been one approach that has gained popularity,15

showing efficacy in terms of reducing agitation,16-18 anxiety17,18 and depression19 and in

improving quality of life.20,21 However, many of these studies have lacked methodological

rigour22 and, thus, more stringently controlled trials are necessary to substantiate claims. The

RCT, considered the ‘gold standard’ of clinical trials23, offers the rigour that many of the

previous studies lack, with the central principles including: randomisation (random

assignment of participants to intervention and control groups); manipulation (manipulation of

the control and intervention); and control (measures taken to reduce the influence of

extraneous variables including a comparable control group).24

In light of the issues outlined above, and the need for more clinical trials in LTC, this

paper provides a timely critical discussion of the challenges encountered whilst conducting a

RCT with cross-over-design. This discussion moves beyond previously published papers by

reflecting on the challenges of using this design (a RCT) to answer research questions in this

setting (LTC) and with this population (older people with dementia) within the one paper. The

challenges critiqued and the ways in which we tried to overcome them are discussed

according to the three principles of a RCT to keep experimental rigour at the forefront of the

discussion and the research area.

STUDY

A RCT with a cross-over design was employed to explore the effect of a live group music

program on agitation, emotion and quality of life of older people with dementia.25-28 The

study followed the recommendations outlined in the Consolidated Standards of Reporting

4

Trials (CONSORT) statement.29 An overview of each main aspect of the study is set out in

Table 1.

[Table 1 here]

CHALLENGES ENCOUNTERED

Randomisation

Sample and Setting

The choice of LTC facility from which to conduct the research is an important first

decision in the research process. The cooperation and flexibility of facility staff is paramount

to the successful implementation of an intervention31 as often there is reluctance to take part

owing to concerns that it may add to current workloads and interrupt ongoing activity

schedules.8,12 In the RCT described here, two Research Assistants (RAs) employed by the

University were based at each of the two facilities during implementation. Their role was to

oversee the logistical aspects of the process, such as managing the transportation of

participants to the intervention. This helped reduce the workload pressures on facility staff but

also ensured that the success of the intervention was not reliant on facility cooperation or

help. This proved especially important in one facility where, in the second half of the

intervention, no facility assistance could be offered to researchers. Budgets are often

exceptionally tight and there is typically only scope for employment of absolutely necessary

personnel. However, it is recommended that the budget should allow for adequate RA support

as this may safeguard against success of the research being dependent on facility assistance

and their level of commitment.

The recruitment process presents a number of challenges to the research team

including identifying and recruiting participants, obtaining consent and the formal enrolment

5

of participants. As in our study, research undertaken within the LTC setting typically sees the

care manager orchestrating the informed consent procedure. This process, however, is often

lengthy and time-consuming, as staff do not always identify participants immediately and do

not send out the consent packs until later still. In the music RCT, the consent process took

around eight – ten weeks and this caused some delay to the start of the intervention. As such,

it is recommended that a generous 12-week consent and recruitment period be allowed for.

Furthermore, if multiple facilities are involved then a longer time-frame may be required as

the speed and success of recruitment can vary, as was our experience.

Further delays to the consent and recruitment process can also be experienced because

the onus, when the person has severe cognitive impairment, is primarily on proxy consent

involving next of kin.31 This can pose challenges as family members can often think that their

relative has health issues that are too severe to allow participation.12 In addition, the next of

kin can sometimes only desire for their relative to take part in the intervention and not the

control activity. This was an issue in our study where families often expressed that their

relative would enjoy the music but not the reading activities. Researchers explained the

importance of the control group in helping to determine the effectiveness of the intervention

and also explained what the control activity would entail. This helped clarify concerns and in

most cases consent was provided. However, this issue is often a challenge in studies involving

a control group and can be an ongoing struggle.

The formal enrolment of participants needs to be carefully considered in terms of

which residents are eligible and their characteristics. In determining the eligibility criteria,

consideration should be given to whether participants from all areas of the LTC facility are to

be included or only those from the Special Care Unit (SCU), high (nursing home) or low

6

(assisted) care respectively. In the music RCT, participants were selected from all areas but

the intervention/control activities were held in central facility locations. For one participant,

movement from SCU to the intervention caused distress and agitation. It may be

advantageous, therefore, for future studies to consider running interventions in separate areas

of the facility so that movement does not differentially influence the measured effectiveness

of the intervention. It is recognised, however, that this would have cost implications by

resulting in the probable need for multiple sites to ensure adequate participant numbers.

Another important consideration is the extent to which participants are screened on

outcome measures prior to formal enrolment. For instance, should there be a minimum score

on outcome measures for study eligibility? In the music project, participants were eligible for

inclusion based on their level of cognitive impairment, as assessed on the Mini Mental State

Examination (MMSE),32 and a documented behavioural history of agitation/aggression on

nursing records within the last month (See Table 1). Despite these pre-requisites, however,

baseline scores on all outcome measures were low. The inclusion of such participants may

mask or attenuate the efficacy of the intervention and, thus, in line with previous

researchers16,23,33 we advocate future studies would benefit from more in-depth screening of

participants prior to study commencement. Although this may compound the difficultly in

finding eligible participants, it may help improve the reliability and validity of findings.

Finally, even though a RCT cross-over design helps reduce the number of participants

to be recruited,34 it may also be advantageous to over-sample when undertaking research with

older people, given that participant attrition is common.8 In the music study, participants were

over-sampled by 10%, and this was accurate for the study timeline and in accounting for the

7

number of deaths (n=5, 10.6%). Future researchers should consider over-sampling by at least

10% when studies are lengthy to maintain study power in the likely event of attrition.

Manipulation

System characteristics of the LTC facility

System characteristics of the LCT facility, such as the schedule of activities and daily routine,

often present challenges in terms of when the intervention can be run.12 Academia often

highlights the need to conduct research when the symptoms under investigation are most

prevalent to maximise the chances of detecting a significant effect.23,33 However, the LTC

facility staff will typically dictate the times when the intervention can be held. For instance, in

the music RCT the two LTC facilities were keen to hold the sessions in the morning, as

activities were planned for the afternoon. This was regardless of when participants’ agitation

levels were at their peak. Researchers, when planning the timing of interventions, should

always take account of the normal facility schedule and ensure it is workable within the

facility. However, it is also worth pursing a discussion regarding the advantages of

conducting the study when individual resident’s symptoms under investigation are at their

peak, as this may increase the chances of discovering a treatment effect and help better

determine if the intervention will be of benefit to participants and care staff.

Logistical issues

When planning the implementation of the intervention and control groups, the location of the

activities must be carefully considered. In our RCT, the music and reading control activities

were held in an activities room in one facility and a chapel in the other. These rooms were

selected by facilities, as no other areas were available, and required the transportation of

participants to the rooms. In addition, the chapel was detached from the main facility

8

buildings and involved a period of transportation outside. This was problematic when weather

conditions were not amenable (i.e., raining) as this reduced willingness to attend sessions. It

would be advisable, therefore, to select rooms that are located within the main facility

building so as to control for such extraneous variables. However, it is recognised that space is

an issue in many facilities and there are often only limited rooms from which to run the

intervention.

The need for consistency during the implementation of the intervention and control is

paramount and this includes promptness of start and end times. However, transporting

participants to the activities in preparation for a punctual start can be difficult. For instance, in

the music program there were some participants who became restless when brought to the

room too early prior to the start of the session. For these participants, the RAs had to ensure

that they were brought just before the sessions commenced, therefore resulting in some

participants waiting for the activity longer than others. In addition, a number of participants

required constant staff supervision, meaning that when brought to the venue, the RA had to

stay with them. This resulted in fewer staff being able to help with the transportation of the

remaining residents. The possible extraneous influence that different lengths of time waiting

for the activity and the different ways that participants’ mood and behaviour can be affected

by transportation to the venue highlights the importance of careful logistical planning, as such

influences need to be controlled in RCTs to ensure validity of findings. It may be

advantageous for community volunteers at the facility to help with the transportation of

residents. One facility in the music program had a bank of volunteers who were able to help

and this was of great assistance.

9

Maintaining session attendance

When undertaking any RCT, there will always be challenges in maintaining participant

session attendance. However, when undertaking research with older people with dementia in

LTC, this can be more of an issue because of the characteristics of the population being

studied. In the music RCT, it was common for participants to refuse to attend a session,

primarily because of ill-health and/or cognitive impairment (i.e., were agitated, confused or

forgot). Some participants also just wanted to attend the music sessions and so refused to

attend any reading activities. Facility staff also contributed to missed sessions by failing to get

participants out of bed or by not asking the participant if they wanted to attend. This

reinforces the importance of having the cooperation of facility staff, as it is beyond the realm

of a research team to ensure that participants are up and dressed ready for the intervention.

Control

Treatment fidelity

Issues associated with treatment fidelity are of integral importance when delivering

intervention and control activities as, ultimately, they can enhance the reliability and validity

of the results.35 At the core of a good treatment fidelity strategy is the standardized procedures

manual.36 A comprehensive manual was produced in our RCT and all staff involved in

delivering the activities were expected to adhere to this manual. Four random spot checks in

the intervention and control activities were undertaken by the research team to ensure that the

sessions were being delivered consistently and in clarifying the role of the musicians and

reading group facilitator (i.e., musicians were to encourage participation such as dancing but

were not to lead behaviour such as getting up to dance first). Also important in maintaining

treatment fidelity was the practical, role-based training in delivering the sessions and in

10

working with older people with dementia. This suite of measures was effective in controlling

the delivering of the sessions and similar steps are recommended for future research.

Outcome measures and data collection

Deciding what instruments and scales to use in a study can be a difficult task, especially given

the wide array available. Four main outcome measures were employed in the live music

program: the Rating Anxiety in Dementia Scale (RAID);37 the Dementia Quality of Life

questionnaire (DQOL);38 the Geriatric Depression Scale (GDS);39 and the Cohen-Mansfield

Agitation Inventory – Short Form (CMAI-SF).40 These measures were chosen because of

their demonstrated reliability and validity but also because they varied in methodological

approach and included proxy reports (care staff completed the CMAI) and self-report (RAs

interviewed the person with dementia on the RAID, DQOL and GDS). By using a range of

outcome measures a more holistic assessment can be sought. This is especially important in

dementia-care research as often, as the severity of dementia develops, verbal communication

can become more difficult.11 It is also useful as it offers insight into any discrepancies on

outcome measures depending on the perspective of the person completing the measure.41

When choosing outcome measures it is also useful to consider using instruments that

are comprised of subscales and not one global, overall score.11 Such scales account for

subtleties in the effect of the intervention. This was the case for the DQOL ‘belonging’

subscale in the music RCT, where a significant effect was detected. If only a global score had

been computed then a Type II error may have occurred.

11

Data analysis and handling missing data

Determining the analysis framework is an important stage in research, as this can determine

the results produced and the findings drawn. In the music study, an ‘Intention-To-Treat’

(ITT) analysis was undertaken; the ‘gold standard’42 as all participants are included

irrespective of the treatment actually received, thus avoiding overestimation of clinical

effectiveness.43,44 When adopting such a framework a second decision needs to be made about

handling missing data. A number of methods can be applied including imputing means,

carrying the last score forward and single and multiple imputations. The last method is

generally regarded as the most sophisticated approach seeing multiple plausible data sets

computed and the results pooled. This was the method chosen for the music study and, as

recommended, a sensitivity analysis was then undertaken to verify the results and confirm no

differences according to approach.43,45 This saw parallel analyses conducted comparing results

when missing data was addressed through multiple imputation, case mean substitutions or left

as missing. As multiple imputation is becoming more common in statistical software, it is

recommended that future researchers consider conducting ITT analysis using this method of

imputation. However, if the use of multiple imputation is deemed appropriate, it should also

be undertaken alongside some sort of sensitivity analysis to verify that the imputation is not

distorting any significant effects.

LESSONS FROM THE FIELD: CLINICAL CONSIDERATIONS

Based upon our experiences of conducting the music RCT, future researchers may benefit

from considering the following questions, outlined in Table 2, when designing and

implementing similar RCTs.

[Table 2 here]

12

CONCLUSION

The characteristics of older people with dementia and the LTC environment can make

conducting research a challenge and, as such, this population subgroup is often understudied,

particularly in terms of clinical or RCTs. In this paper we have critically discussed some of

the difficulties that we faced whilst implementing a RCT involving a live music program with

older people with dementia in LTC. It is hoped that this discussion provides a starting point

for researchers designing similar RCTs and helps encourage a proactive approach in dealing

with research-related challenges right from project conception.

ACKNOWLEDGEMENTS

This paper reports on a larger study funded by the National Health & Medical Research

Council, Australia (Grant ID 481929). The involvement of the sponsor was monetary and did

not include support in the design, collection and analysis of data or the final report. The

authors acknowledge support and contributions by RSL care staff, family and residents during

the research process.

13

REFERENCES

1 Australian Institute of Health and Welfare [AIHW]. Dementia in Australia:

national data analysis and development. Canberra: 2007. Australian Institute

of Health and Welfare, AIHW Cat No. AGE 53.

2 Boller FM, Verny M, Hugonot-Diener L, Saxton J. Clinical features and

assessment of severe dementia: a review. European Journal of Neurology

2002; 9: 125-136.

3 Ory MG, Hoffman RR III, Yee JJ, Tennstedt S, Schulz R. Prevalence and

impact of care giving: a detailed comparison between dementia and

nondementia caregivers. The Gerontologist 1999; 39: 177-185.

4 Mott S, Kingsley B. Challenging behaviour: the issues of quality of care. In:

Garratt RNS (eds), Nursing Older People: Issues and innovations. Sydney:

MacLennan & Petty Pty Ltd, 1999; 212-233.

5 Goergen T. Stress, conflict, elder abuse and neglect in German nursing homes:

a pilot study amongst professional caregivers. Journal of Elder Abuse 2001;

13: 1-26.

6 Ferri CP, Prince M, Brayne C et al. Global prevalence of dementia: a Delphi

consensus study. Lancet 2005; 366: 2112-2117.

14

7 Hall S, Longhurst S, Higginson IJ. Challenges to conducting research with

older people living in nursing homes. BMC Geriatrics 2009; 9:(38).

8 Maas ML, Kelley LS, Park M, Specht JP. Issues in conducting research in

nursing homes. Western Journal of Nursing Research 2002; 24: 373-389.

9 Mentes JC, Tripp-Reimer T. Barriers and facilitators in nursing home

intervention research. Western Journal of Nursing Research 2002; 24: 918-

936.

10 Snyder M, Tseng Y, Brandt C et al. Challenges of implementing intervention

research in persons with dementia: example of a glider swing intervention.

American Journal of Alzheimer’s Disease and Other Dementias 2001; 16: 51-

56.

11 Mittelman MS. Psychosocial intervention research: challenges, strategies and

measurement issues. Aging & Mental Health 2008; 12: 1-4.

12 Cohen-Mansfield J. Nonpharmacologic interventions for inappropriate

behaviours in dementia: a review, summary and critique. American Journal of

Geriatric Psychiatry 2001; 9: 361-381.

13 Working group on functional outcome measures for clinical trials. Functional

outcomes for clinical trials in frail older persons: time to be moving. Journal

of Gerontology: Medical Sciences 2008; 63A: 160-164.

15

14 Jackson J, Mandel D, Blanchard J et al. Confronting challenges in intervention

research with ethically diverse older adults: the USC Well Elderly II Trial.

Clinical Trials 2009; 6: 90-101.

15 Bruer RA, Spitznagel E, Cloninger CR. The temporal limits of cognitive

change from music therapy in elderly persons with dementia or dementia-like

cognitive impairment: a randomised controlled trial. Journal of Music Therapy

2007; 44: 308-328.

16 Ledger AJ, Baker FA. An investigation of long-term effects of group music

therapy on agitation levels of people with Alzheimer’s Disease. Aging &

Mental Health 2007; 11: 330-338.

17 Raglio A, Bellelli G, Traficante D et al. Efficacy of music therapy in the

treatment of behavioural and psychiatric symptoms of dementia. Alzheimer’s

Disease and Associated Disorders 2008; 22: 158-162.

18 Svansdottri HB, Snaedal J. Music therapy in moderate and severe dementia of

Alzheimer’s type: a case-control study. International Psychogeriatrics 2006;

18: 613-621.

19 Ashida S. The effect of reminiscence music therapy sessions on changes in

depressive symptoms in elderly persons with dementia. Journal of Music

Therapy 2000; 37: 170-182.

16

20 Clair AA, Bernstein B. A preliminary study of music therapy programming for

severely regressed persons with Alzheimer’s-type dementia. The Journal of

Applied Gerontology 1990; 9: 299-311.

21 Rio R. Improvisation with the elderly: moving from creative activities to

process-oriented therapy. The Arts in Psychotherapy 2002; 29: 191-201.

22 Boso M, Politi P, Barale F, Emanuele E. Neurophysiology and neurobiology

of the musical experience. Functional Neurology 2006; 21: 187-191.

23 O’Connor DW, Ames D, Gardner B, King M. Psychosocial treatments of

behaviour symptoms in dementia: a systematic review of reports meetings

quality standards. International Psychogeriatrics 2009; 21: 225-240.

24 Taylor B, Kermode S, Roberts K. Research in nursing and health care:

Evidence for practice (3rd ed.). Victoria: Thomson, 2006.

25 Cooke M, Moyle W, Shum D, Harrison S, Murfield J. The effect of music on

quality of life and depression in older people with dementia: a randomised

control trial. Alzheimer’s & Dementia 2009; 5: e11.

26 Cooke M, Moyle W, Shum D, Harrison S, Murfield J. A randomized

controlled trial exploring the effect of music on agitated behaviours and

anxiety in older people with dementia. Aging and Mental Health in press.

17

27 Cooke M, Moyle W, Shum D, Harrison S, Murfield J. A randomised

controlled trial exploring the effect of music on quality of life and depression

in older people with dementia. Journal of Health Psychology in press.

28 Harrison S, Cooke M, Moyle W, Shum D, Murfield J. Development of a

music intervention protocol and its effect on participant engagement:

experiences from a randomised controlled trial with older people with

dementia. Arts & Health in press.

29 Schulz KF, Altman DG, Moher D. CONSORT 2010 Statement: updated

guidelines for reporting parallel group randomised trials. Annals of Internal

Medicine 2010; 152:1-7.

30 American Psychiatric Association. Diagnostic and Statistical Manual of

Mental Disorders (4th ed.). USA: American Psychiatric Association, 1994.

31 Finkel S, Cooler C. Clinical experiences and methodologic challenges in

conducting clinical trials on the behavioural disturbances of dementia.

International Psychogeriatrics 1996; 8:151-164.

32 Folstein M, Folstein S, McHugh P. Mini-mental state. Journal of Psychiatric

Research 1975; 12: 189-198.

18

33 O’Connor DW, Ames D, Gardner B, King M. Psychosocial treatments of

psychological symptoms in dementia: a systematic review of reports meetings

quality standards. International Psychogeriatrics 2009; 21:241-251.

34 Senn S. Cross-over trials in clinical research (2nd ed.). Chichester: Wiley,

2002.

35 Resnick B, Ignuito P, Orwig D et al. Treatment fidelity in behaviour change

research: a case example. Nursing Research 2005; 54:139-145.

36 Chambless D, Hollon SD. Defining empirically supported therapies. Journal

of Consulting and Clinical Psychology 1998; 66: 7-18.

37 Shankar KK, Walker M, Frost D, Orrell MW. The development of a valid and

reliable scale for rating anxiety in dementia (RAID). Aging and Mental Health

1999; 3: 39-49.

38 Brod B, Stewart A, Sands L, Walton P. Conceptualization and measurement of

quality of life in dementia: dementia quality of life instrument (DQoL). The

Gerontologist1999; 39: 25-35.

39 Yesavage JA, Brink TL, Rose TL et al. Development and validation of a

geriatric depression screening scale: a preliminary report. Journal of

Psychiatric Research 1983; 17: 37-49.

19

40 Werner P, Cohen-Mansfield J, Koroknay V, Braun J. The impact of a

restraint-reduction program on nursing home residents. Geriatric Nursing

1994; 15:142-146.

41 Cotrell V, Schulz R. The perspective of the patient with Alzheimer’s Disease:

a neglected dimension of dementia research. The Gerontologist 1993; 33: 205-

211.

42 Heritier S, Gebski V, Keech A. Inclusion of patients in clinical trial analysis:

the intention-to-treat principle. The Medical Journal of Australia 2003; 179:

438-440.

43 Hollis S, Campbell F. What is meant by intention to treat analysis? survey of

published randomised controlled trials. British Medical Journal 1999; 319:

670-674.

44 Kruse RL, Alper BS, Reust C, Stevermer JJ, Shannon S, Williams R.

Intention-to-treat analysis: Who is in? Who is out? The Journal of Family

Practice 2002; 51: 969-971.

45 Sterne J, White I, Carlin J et al. Multiple imputation for missing data in

epidemiological and clinical research: potential and pitfalls. British Medical

Journal 2009; 339: 157-160.

20

Table 1 Overview of study

Methodological aspect

Description of study

Design Randomised cross-over design with music intervention and reading control group, from October 2008 to March 2009.

Ethical approval granted by University human research ethics committee and a support statement provided by partner

aged care organisation.

Sample 69 assessed for eligibility and 47 formally enrolled into the research based on criteria: 1. a confirmed diagnosis of early

to mid stage dementia OR probable dementia (i.e., a cognitive impairment level of 12-24 on MMSE) OR features

consistent with dementia of Alzheimer’s type as per DSM–IV30 AND 2. a documented behavioural history of

agitation/aggression on nursing/medical records within the last month. Consent provided by next of kin and, where

possible, participants. The study’s biostatistician, blinded to participant identity, undertook randomisation process of

treatment group allocation.

Setting Two mixed-gender LTC facilities North of Brisbane, Australia. Site A = 164 residents and Site B = 94 residents.

Intervention Intervention and control activities ran for 40-minutes, 3 mornings a week for 8 weeks. After a 5-week ‘washout’ period,

participants crossed-over into opposite activity and protocol repeated for another 8 weeks. Intervention was a live group

music program involving 30-minutes of musician-led familiar song singing and 10-minutes of pre-recorded

21

instrumental music. Reading control sessions included reading local news stories, short stories, telling jokes and quiz

activities.

Data collection Assessments at baseline, mid-point (wash-out period) and post-intervention on 4 outcome measures:

By RAs: RAID; DQOL; and GDS.

By care staff: CMAI-SF.

Data analysis Data entered and analysed using the Statistical Package for the Social Sciences Version 17.0 (SPSS Inc., Chicago, IL,

USA). Following a missing values analysis, an ITT framework was undertaken and missing values in outcome

measures were addressed through a multiple imputation method.

22

Table 2 Clinical considerations when planning a RCT with a psychosocial intervention involving older people with dementia in LTC Randomisation Is the facility in a position to support the

research (i.e., assessment of staffing levels;

enthusiasm for the research)?

Can the budget allow for the provision of

some RA level support for implementing the

intervention?

Can the project timings allow for a 12-

week recruitment process?

Are multiple sites being used? If so, is

there flexibility to extend the recruitment

process further (i.e., 14-week)?

Does the research require participants from

all areas of the facility or just one area? (i.e.,

high, low and/or special care)?

Manipulation What activities are fixed for the facility

and need to be worked around (i.e. morning

tea, lunch etc)?

Can the interventions be run when

symptoms under investigation are most

prevalent?

Is the suggested room where activities are

to be held detached from the main building

and does getting to it involve going outside?

Which participants should be transported

to the intervention first (i.e., who cannot be

left unattended; who becomes agitated if left

waiting etc)?

What equipment is needed and are special

Control What measures are going to be employed

to ensure treatment fidelity (i.e., standardised

procedure; practice intervention sessions;

training; spot checks etc)?

Which outcome measures will be

appropriate and what is their reliability and

validity?

Can/should outcome measures with

subscales be used to ensure subtleties in effect

are captured?

How will interrater reliability be controlled

(i.e., training, blinding etc)?

Can an ITT analysis framework be adopted

(i.e., is data Missing At Random)?

23

Is screening/assessment of participants on

an outcome measure necessary (i.e., is a

minimum level/score required for eligibility)?

requirements necessary (i.e., ease of

re/assembly, level of noise etc)?

Are staff aware of which residents are

taking part in which activity and the need for

them to be ready if they wish to attend?

Can time for a sensitivity analysis of data

be built into the project?